Pharmacology and Physiology of Central Noradrenergic Systems

Stephen L. Foote and Gary S. Aston-Jones

The purpose of this chapter is to summarize and critically evaluate selected major recent developments in our understanding of the cellular physiology and pharmacology of brain noradrenergic neurons. There has been substantial progress in these areas since the previous edition of this volume. We have chosen to focus on particular subsets of data that form "critical masses" of information relevant to prominent functional issues. These include data obtained using both in vitro and in vivo preparations to accomplish detailed studies of the noradrenergic neurons of the locus coeruleus (LC) as well as their target cells. Those findings having clear implications for understanding systems-level physiology and pharmacology, as well as those findings with readily apparent behavioral implications, have been emphasized.

Recordings obtained from LC neurons in brain slices have proved enormously valuable in characterizing the pharmacologic properties of these cells. Such in vitro preparations have numerous advantages for pharmacological studies of CNS neurons (reviewed in Electrophysiology).

The LC slice has been used to greatest advantage in the study of two prominent receptors on this cell population, the alpha-2 adrenoceptor and the mu opiate receptor. Because of space constraints, the present summary of recent data concerning LC pharmacology as studied in the slice preparation is focused on these receptors. The reader is also referred to numerous reports in the literature (11, 12, 13, 45, 48, 49, 50, 51, 55, 64) for descriptions of other pharmacologic properties of LC neurons in the slice.

Early in vivo studies by Aghajanian and colleagues provided evidence for potent inhibition of LC neurons by alpha-2 adrenergic agonists, such as clonidine (reviewed in ref. 25). In vivo studies indicated that one possible role of these receptors is autoinhibition of LC neurons via recurrent collaterals (12, 25). While these extracellular studies indicated that norepinephrine (NE) and epinephrine strongly suppressed LC impulse activity, and that alpha-2 receptors were the most prominent adrenoceptor on LC cells, they revealed little about the underlying membrane effects of these agents on LC neurons.

Intracellular studies in LC slices revealed important aspects of the cellular events triggered by alpha-2 adrenoceptor activation. Egan et al. (23) found that electrical field stimulation of the slice yielded potent hyperpolarizing synaptic potentials in LC neurons that were mediated by alpha-2 receptors, consistent with the proposed alpha-2 mediation of collateral inhibition. Other studies (5) questioned the role of alpha-2 receptors in autoinhibition; however, this conflict with results of previous in vivo studies from the same laboratory may reflect the limiting effects of severed axonal or dendritic collaterals in the slice preparation.

Williams et al. (60) used voltage-clamp analysis in slices to show that alpha-2-mediated hyperpolarization of LC neurons resulted from opening an inwardly rectifying potassium channel. Williams and North (62) also found that NE acting at alpha-2 receptors on LC neurons reduced calcium influx into these cells, which contributed to the inhibition of discharge. Subsequent studies (2, 44) showed that the alpha-2-induced increase in potassium conductance was mediated through a G-protein intermediary and caused a decrease in adenylate cyclase activity and intracellular cAMP levels (reviewed in Signal Transduction Pathways for Catecholamine Receptors and Intracellular Messenger Pathways as Mediators of Neural Plasticity). Interestingly, however, the alpha-2-evoked opening of K channels and subsequent hyperpolarization was not mimicked by forskolin (an activator of adenylate cyclase) or blocked by inhibitors of the cAMP-protein kinase A pathway (43). These observations indicate that the potassium channel is linked to the alpha-2 receptor directly by a G protein without an intermediary diffusible second messenger, such as cAMP. The role of cAMP, and its regulation by alpha-2 adrenoceptor activity, remains uncertain and controversial (6, 44).

Together, the above studies yield a clear picture of the action of adrenergic agonists on LC neurons. Acting almost entirely through alpha-2 receptors (at least in adults; see refs. 41 and 61, for differences in LC pharmacologic properties in young rats), NE or epinephrine increases an inwardly rectifying potassium conductance. It appears that the alpha-2 receptor is linked to a G-protein mechanism (perhaps within the membrane) and that the G protein is directly linked to the K channel which it opens without the assistance of a diffusible second messenger. The details of alpha-2 adrenoceptor action are probably better understood for LC neurons than any other class of central neuron. However, other brain areas that have been examined with intracellular recordings in slice preparations have exhibited similar membrane responses, indicating that the LC slice may serve as a generalizable model system for alpha-2 receptor actions.

LC neurons in rat are densely invested with opiate receptors, particularly of the mu subtype (25). This nucleus is also prominently innervated by endogenous opioid fibers (13). Early studies using extracellular recordings in vivo demonstrated that systemic or iontophoretic opiates strongly inhibited LC discharge activity (reviewed in ref. 25). These findings, and the proposed role of the LC system in opiate abuse (see Signal Transduction Pathways for Catecholamine Receptors, Physiological and Anatomical Determinants of Locus Coeruleus Discharge: Behavior and Clinical Implications, and Intracellular Messenger Pathways as Mediators of Neural Plasticity), motivated a series of elegant studies of the effects of opiates on membrane properties of LC neurons.

Intracellular studies in slice preparations by Williams, North, and colleagues first showed that enkephalin acted at mu receptors on LC neurons to increase a K conductance and hyperpolarize the membrane (reviewed in ref. 11). Later studies (42) showed that this action of enkephalin also inhibited calcium action potentials in LC neurons, a second mechanism (besides hyperpolarization) by which opioids may inhibit LC discharge activity. Additional experiments determined that the potassium conductance activated by opioids in LC neurons was inwardly rectifying and operated through a G-protein mechanism (2, 44). Biochemical studies revealed that acute opiate treatment decreased adenylate cyclase activity and cAMP levels in LC neurons (see Signal Transduction Pathways for Catecholamine Receptors and Intracellular Messenger Pathways as Mediators of Neural Plasticity). Some studies indicated that the hyperpolarization induced by opiates in LC cells was linked to the reduction of cAMP, as it could be attenuated by administration of certain cAMP analogues to the slice bath (6). However, patch-clamp experiments by Miyake et al. (39) showed that opioids potently activate a potassium conductance in LC neurons in the absence of any intermediary diffusible second messenger such as cAMP. The role of cAMP in responses of LC neurons to opioids remains unclear. However, recent studies have demonstrated that increases in cAMP in LC neurons (e.g., as brought about by other transmitter inputs) markedly increase the hyperpolarization evoked in LC neurons by opiates (reviewed in ref. 11).

Examination of the properties listed above for alpha-2 adrenoceptor-and mu opiate receptor-mediated actions in LC neurons reveals striking similarities: Both open an inwardly rectifying potassium channel through a G-protein mechanism, both inhibit calcium influx into LC neurons, and both decrease cAMP levels in these cells. Many of these common features were shown to reflect the fact that alpha-2 and mu receptors are linked to the same potassium channel (2, 44). The main evidence for this intriguing finding is that during a maximal electrophysiological response to an agonist at one receptor, application of an agonist at the other receptor does not further increase K conductance. This and other evidence indicate that the two receptors share the same K channels. This is one of the first and best characterized examples of shared channels among different receptors in central nervous system (CNS) neurons, but other examples also exist.

The LC has also been used as a model system to study potential cellular mechanisms underlying opioid tolerance and dependence. It has been known from early in vivo studies that LC neurons become tolerant to the inhibitory effects of opiates with chronic administration, and that opiate withdrawal potently activates LC neurons (reviewed in ref. 11). Detailed examination of changes in LC response to opiates using intracellular recordings in vitro revealed that tolerance may be due primarily to a decrease in the coupling of the mu receptor to the G protein, which, in turn, activates the K channel (17). The mechanism of withdrawal hyperactivity revealed some interesting and unexpected twists. First, it was found that although LC neurons in slices of morphine-treated rats exhibited tolerance, they exhibited little dependence; that is, the withdrawal-induced hyperactivity typical of LC neurons in vivo was relatively lacking when examined in the in vitro slice (reviewed in Physiological and Anatomical Determinants of Locus Coeruleus Discharge: Behavior and Clinical Implications, as well as in ref. 11). This indicated that most of the opiate dependence of LC neurons was due to a change in afferent drive to these cells rather than a change in the membrane properties of the cells themselves. This was supported by several findings. First, intracellular studies showed little, if any, change in intrinsic membrane properties of LC neurons in slices taken from chronically morphine-treated rats (17). Second, excitatory amino acid antagonists, administered either intracerebroventricularly or locally into the LC in vivo, substantially attenuated the hyperactivity elicited by morphine withdrawal in intact, morphine-pretreated rats (). Finally, lesions of the nucleus paragigantocellularis, a prominent afferent to the LC that utilizes an excitatory amino acid transmitter to activate LC neurons, also attenuated withdrawal-induced hyperactivity of these cells. Together, these results strongly indicate that the bulk of opiate withdrawal response in the LC is mediated via augmented amino acid drive to the LC from the nucleus paragigantocellularis.

These findings, in view of previous results showing that serotonin decreases LC response to excitatory amino acids () (7), suggested the clinically interesting possibility that serotonergic drugs may attenuate LC hyperactivity during opiate withdrawal. Indeed, Akaoka and Aston-Jones (4) have recently reported that fenfluramine, fluoxetine, or sertraline have this effect, suggesting that they may be useful clinically in treating withdrawal () (see Physiological and Anatomical Determinants of Locus Coeruleus Discharge: Behavior and Clinical Implications).

It should also be noted that while most of the withdrawal response can be accounted for by extrinsic afferent drive, recent studies reveal that a small amount of the withdrawal response (~20%) may reflect local intracoerulear changes that may involve alterations in adenylate cyclase activity (see Signal Transduction Pathways for Catecholamine Receptors and Intracellular Messenger Pathways as Mediators of Neural Plasticity).

It has long been hypothesized that a principal function of the LC is participation in the control of various stages of the sleep–waking cycle (reviewed in ref. 9). In rat, spontaneous LC discharge covaries consistently with stages of the sleep–waking cycle: These neurons fire fastest during waking, more slowly during slow-wave sleep, and become virtually silent during paradoxical sleep (PS) (8). These observations support previous proposals that a similar subpopulation of neurons within the neurochemically heterogeneous cat LC are noradrenergic (8). However, other activity profiles of purported noradrenergic neurons have been reported in cat LC (18). Similar fluctuations in LC discharge with levels of alertness are also seen in the monkey (8). Although these animals do not exhibit normal sleep and waking under our experimental conditions of chair restraint, LC activity has been observed during alertness and drowsiness as measured by electroencephalography (EEG). As described above for rat LC, monkey LC neurons vary their activity closely with the state of arousal, even during unambiguous waking. Thus, periods of drowsiness are accompanied by decreased LC discharge, whereas alertness is consistently associated with elevated LC activity. Also as in rat, such changes in LC activity preceded the corresponding changes in EEG state by a few hundred to several hundred milliseconds.

It has further been observed that LC discharge is altered during certain spontaneous waking behaviors. During both grooming and consumption of a glucose solution, rat LC discharge decreases compared to that in other behavioral episodes characterized by approximately the same degree of EEG arousal (8). Similar results have been obtained for LC activity in behaving primates. These results indicate that LC discharge is reduced not only for periods of low arousal (drowsiness or sleep), but also during certain behaviors (grooming and consumption) during which animals are in an active waking state but are inattentive to most environmental stimuli.

LC discharge rates also covary reliably with orienting behavior. In both rats and monkeys, the highest discharge rates observed for LC neurons were consistently associated with spontaneous or evoked behavioral orienting responses (8). LC discharge associated with orienting behavior is phasically most intense when automatic, tonic behaviors (sleep, grooming, or consumption) are suddenly disrupted and the animal orients toward the external environment.

In addition to the above fluctuations in LC spontaneous discharge, it has also been observed that these neurons in unanesthetized rats and monkeys are responsive to non-noxious environmental stimuli (8). In waking rats, LC activity is markedly phasic, yielding short-latency (15–50 msec) responses to simple stimuli in every modality tested (auditory, visual, somatosensory, and olfactory). Responses were most consistently evoked by intense, conspicuous stimuli, though sporadic responses were also observed for nonconspicuous stimuli as well. These responses were similar for the different sensory modalities, and they consisted of a brief excitation followed by diminished activity lasting a few hundred milliseconds.

While sensory responsiveness was qualitatively similar for LC neurons in rats and monkeys, there were important differences as well. In rats, any of a variety of intense stimuli evoked LC responses in a majority of sensory trials. In contrast, monkey LC was less strongly influenced by such stimuli, with responses fading after the first few trials. However, more complex stimuli, such as a new face or a meaningful but unpredictable stimulus (see below), were consistently capable of eliciting LC responses in monkeys (8). In both species, stimuli that interrupted behavior and elicited orienting responses were those that most reliably evoked LC responses.

The results described above indicated that LC neurons are robustly activated by intense stimuli because their intensity elicited behavioral orienting responses. However, it was hypothesized that nonintense stimuli that demand a behavioral response may also elicit responses in LC neurons. To explicitly test this possibility, Aston-Jones et al. (8, 10) have recorded LC activity in unanesthetized monkeys trained in an "oddball" visual discrimination task. This task involves discriminating different colored lights, or vertical versus horizontal line segments, on a video monitor. Target cues (CS+) are presented on 10–20% of trials, intermixed in a semirandom fashion with non-target cues (CS-). Neurons in the LC area were recorded along with cortical surface slow waves [averaged event-related potentials (AERPs)] and behavioral responses (hits, misses, false alarms, and correct rejections). LC neurons exhibited phasic as well as tonic activity during this task, an observation that links this system to attentional processing.

In terms of phasic responses, LC neurons consistently and uniformly were activated by CS+ stimuli but not by other task events. In addition, these phasic responses to CS+ stimuli occurred with a short latency (mean 108 msec). This is far in advance of behavioral responses which occur at 250–300 msec, indicating that LC responses may participate in (e.g., facilitate) the behavioral response to CS+ cues. Recordings during reversal training revealed that these responses were specifically related to the imperative nature of stimuli, not to their physical attributes. As illustrated in , when the stimulus meaning was reversed, LC neurons quickly reversed their stimulus responsiveness, so that responses were soon selectively elicited for the new CS+ (previous CS-) while responses for the old CS+ (new CS-) rapidly faded. It is noteworthy that these changes varied closely with behavioral performance, so that responses to the new CS+ increased (and responses to the new CS- decreased) in parallel with the increasing percentage of correct behavioral responses to the new contingency.

In addition, cortical activity exhibited a similar set of properties. AERPs recorded from the frontal and parietal cortices at latencies of 200–300 msec post-stimulation were selectively augmented by CS+ cues (8), as reported by others in both humans (31) and monkeys (46). There is evidence that these potentials in nonhuman primates are similar to the P3 or P300 potentials in humans (46). During reversal training, the AERPs altered their selectivity in a manner similar to that of neurons in the LC area to become selectively responsive to the new CS+ and no longer responsive to the previous CS+ (new CS-) (8). As with the brainstem neurons, these changes in cortical evoked activity followed a time course that closely paralleled behavioral discrimination performance during reversal. Pineda et al. (46) have shown that such AERP responses in monkeys are attenuated by lesions of the LC. The present results support their suggestion that the LC may contribute to these cortical slow-wave events.

Tonic activity of monkey LC neurons also changed in an intriguing manner during this task (47). In one version of the task, the animal must visually fixate a spot on the video monitor for a few hundred milliseconds to initiate each trial. Such foveation is effortful and reflects focused attentiveness to the task. As noted above, during drowsiness LC activity is very low (<0.5 spike/sec) and there is typically no task performance. It was observed that during continuous alertness and task performance the frequencies of both LC discharge and successful foveation fluctuated over short (10–30 sec) and long time intervals (10–30 min). Periods of high foveation frequency were associated with (a) stable gaze directed at the center of the monitor and (b) good overall task performance. Epochs of low foveation frequency, in contrast, were associated with frequent eye movements and poor overall performance. Changes in fixation frequency and task performance were consistently inversely correlated with LC discharge rate, such that slightly elevated LC activity (by about 1–2 spikes/sec) was accompanied by decreased foveation frequency and poorer task performance. Correlation analyses revealed that this relationship was highly significant (typically r = -0.4, p < 0.001). Thus, a strong relationship exists among tonic LC discharge rate, sensory responsiveness of LC neurons, and vigilance performance. Very low LC activity is associated with drowsiness and inattentiveness, whereas high tonic LC discharge corresponds with labile scanning attention; optimal focusing of attention occurs with intermediate levels of tonic LC activity (47; see also Central Norepinephrine Neurons and Behavior and Physiological and Anatomical Determinants of Locus Coeruleus Discharge: Behavior and Clinical Implications).

The substantial involvement of the LC, along with enhanced NE release, in stress responses has been further documented and elaborated in recent years (see Valentino and Aston-Jones).

Three techniques have commonly been used to assess the impact of LC on the electrophysiological activity of its target neurons: (i) in vivo microiontophoretic or pressure application of NE (and/or agonists or antagonists), (ii) in vivo LC activation (or inactivation), and (iii) in vitro application of NE and related agents (see Electrophysiology).

There have been numerous studies focused on the issue of how NE and/or LC stimulation alter the spontaneous and sensory-evoked discharge patterns of neocortical neurons (see refs. 25, 26, and 63) for reviews). These studies have yielded generally consistent findings indicating that these manipulations are capable of enhancing the responsiveness of these neurons to sensory stimulation while at the same time reducing or not altering spontaneous activity. In recent years, Waterhouse and colleagues (e.g., see refs. 40, 58, and 59) have extended and refined this body of data, examining the ability of NE application and/or LC stimulation to "gate" inputs to target neurons so that previously subthreshold synaptic input becomes suprathreshold for eliciting discharge activity. As in their previous work, these investigators find evidence for NE enhancement of both excitatory and inhibitory inputs to neocortical target cells. For example, the effects of iontophoretically applied NE on visually elicited responses of neurons in area 17 of anesthetized rat have been examined (58), 59). For the majority of cells tested, NE application was found to enhance both the vigor and the precision of visually evoked responses.

Within the last decade, there have been numerous demonstrations obtained from both in vitro and in vivo preparations that exogenous NE and/or LC stimulation can substantially enhance various measures of synaptically driven neuronal responses in the hippocampus. The early demonstrations of such phenomena followed the pioneering studies of Madison and Nicoll (33) demonstrating that NE reduces spike frequency adaptation, elicited by current injection, via actions on specific conductances in individual hippocampal neurons. These earlier studies, not summarized here, demonstrated facilitatory NE effects on various types of synaptically elicited activity in hippocampus (reviewed in ref. 27). Recent studies have confirmed and extended these previous observations. For example, both exogenous NE and LC stimulation have been shown to increase the amplitude of the population spike elicited in the dentate gyrus by perforant path stimulation, and it has been demonstrated that this enhancement can be blocked by beta-receptor antagonists and reduced by NE depletion (e.g., see refs. 20, 28, 29, 52, and 57; see ref. 29 for limitations on the effects of beta blockers). It has also been recently shown that this beta-dependent potentiation can be elicited by stimulation of a major LC afferent, the nucleus paragigantocellularis (14). A long-lasting, beta-mediated enhancement of the population spike evoked in CA1 by Schaffer collateral stimulation has also been documented (22, 30). In addition to these effects, which can be elicited without previous high-frequency stimulation of the non-NE afferent, there have also been numerous demonstrations of NE enhancement of long-term potentiation in CA3 and the dentate gyrus (28, 32, 52).

Thus, the literature demonstrating LC/NE enhancement of synaptically driven activity in hippocampus is impressive in several regards: Compatible observations have been made in vivo and in vitro; NE application and LC stimulation produce similar effects; several laboratories have replicated the basic findings, indicating that they are robust and consistent; and similar effects have been demonstrated for various hippocampal synapses.

One of the most significant advances in understanding the possible mechanisms underlying the functional impact of the LC–NE system on behavioral state has been the delineation of the profound effects of exogenous NE on patterns of neuronal discharge activity in thalamus and neocortex in in vitro brain-slice preparations. A cohesive set of findings has been generated by McCormick and his colleagues (e.g., see refs. 34, 35, 36, 37, 38, and 56), who have obtained intracellular recordings from neurons in numerous thalamic and cortical areas in such preparations. These experiments show that NE activates these cells and changes their discharge patterns from one that in the in vivo preparation is generally observed during slow-wave sleep to one that is characteristic of the EEG state that indexes alertness and arousal. In addition, these investigators have been able to specify the membrane events underlying these pattern changes. For example, in studies of neurons from several thalamic nuclei, NE induced a slow depolarization that was apparently due to a decrease in potassium conductance (35, 37, 38). This slow depolarization suppressed burst firing and enhanced single-spike activity, changes in discharge pattern that mimic those that occur in these neurons in vivo during the transition from slow-wave sleep to waking and are presumed to underlie the parallel changes observed in corticothalamic EEG indices. Thus, these in vitro studies have provided evidence that a cellular substrate exists that could well serve to mediate profound LC effects on behavioral state as measured by thalamocortical EEG measures.

Recently, in vivo experiments have been conducted in which LC activity has been manipulated via local drug infusions while being simultaneously monitored with microelectrode recordings. This method achieves selective, acute, potent, and verifiable activation of LC, using a combined recording/infusion probe, as previously described (1, 24). The electrophysiological recordings facilitate the accurate placement of peri-coerulear infusions of drugs that alter LC neuronal discharge rates. The close proximity of the infusion site to the LC allows the use of small volumes, reducing the spread of the infusion into other brainstem structures while at the same time not damaging the LC itself. The microelectrode recordings obtained before and after the infusion provide quantitative verification of the LC manipulation and permit analyses of temporal relationships between the onset and offset of LC activation/inactivation and any observed physiological effects.

Recently, this technique has been used to determine whether peri-LC bethanechol infusions produce reliable forebrain EEG activation, whether this EEG activation is dependent on enhancement of LC neuronal discharge rates, and whether this effect can be blocked by antagonizing noradrenergic neurotransmission. In a total of 39 halothane-anesthetized animals, the findings were as follows (15); examples are shown in : (a) LC activation was consistently followed, within 5–30 sec, by a shift from low-frequency, high-amplitude to high-frequency, low-amplitude activity in the neocortical EEG; (b) these EEG responses followed LC activation with similar latencies whether infusions were made lateral or medial to the LC; (c) infusions placed at a distance of more than 500–600 mm from the LC were not followed by these EEG responses; (d) following infusion-induced activation, EEG returned to preinfusion patterns with about the same time course as the recovery of LC activity; and (e) the infusion-induced changes in EEG were blocked or severely attenuated by pretreatment with the alpha-2 agonist clonidine (50 mg/kg, iv) or the beta-antagonist propranolol (200 mg, icv). These observations indicate that enhanced LC discharge activity is the crucial mediating event for the infusion-induced changes in forebrain EEG activity observed under these conditions. This demonstration that LC activation is followed by cortical EEG desynchronization is especially interesting because it indicates that LC activity levels are not only correlated with, but can be causally related to, EEG measures of forebrain activation.

It is well known that systemic administration of alpha-2 adrenergic agonists produces behavioral and EEG measures of sedation. Because these drugs act to inhibit LC neuronal discharge activity and NE release (see preceding discussion), these observations are consistent with an action of the LC–NE system in the maintenance of an activated forebrain. The possibility that inhibition of LC activity might be a major mediating mechanism for these actions has been enhanced by the observation that intrabrainstem administration of alpha-2 agonists into the region of the LC has similar sedative effects (19, 21). However, interpretation of these results is complicated by a variety of factors, such as the small size of the LC and its close proximity to other nuclei known to affect behavioral and EEG states (53, 54). These factors, together with the absence of electrophysiological measures documenting the relationship between changes in LC neuronal activity and EEG state following such infusions, preclude specific conclusions regarding the site(s) of action for the sedative effects of intrabrainstem administered alpha-2 agonists.

In recent studies (16), small clonidine infusions (35 nl or 150 nl; 1 ng/nl) were made either immediately adjacent to, or at a distance of approximately 1000 mm from, LC in halothane-anesthetized rats using a recording/infusion probe, as described above. These infusions were made under conditions in which high-frequency, low-voltage activity predominated in the neocortical EEG. The following results were obtained (see for examples): (a) cortical EEG activity was not substantially affected following unilateral clonidine-induced LC inactivation; (b) bilateral clonidine infusions that completely suppressed LC neuronal discharge activity in both hemispheres induced a shift in neocortical EEG to low-frequency, large-amplitude activity; (c) 35-nl infusions placed 800–1200 mm from the LC did not induce a complete suppression of LC activity and did not alter forebrain EEG; (d) 150-nl infusions placed 800–1200 mm from LC were either ineffective at completely suppressing LC neuronal discharge activity or did so with a longer latency to complete LC inhibition and a shorter duration of inhibition; (e) in all cases, the onsets of EEG responses coincided with the complete bilateral inhibition of LC discharge activity, and these EEG effects persisted throughout the period during which bilateral LC neuronal discharge activity was completely suppressed (60–240 min); and (f) the resumption of preinfusion EEG activity patterns closely followed the recovery of LC neuronal activity or could be induced with systemic administration of the alpha-2 noradrenergic antagonist, idazoxan. These results suggest that the clonidine-induced EEG changes were dependent on the complete bilateral suppression of LC discharge activity and that, under the present experimental conditions, the LC–NA system exerts a potent and tonic activating influence on forebrain EEG state such that activity within this system is necessary for the maintenance of an activated forebrain EEG state.

In these studies of neocortical EEG and manipulation of LC activity by drug infusion, hippocampal EEG activity was also recorded. It was consistently observed that LC activation was followed by intense hippocampal theta activity, whereas LC inactivation diminished the occurrence of EEG activity in this frequency range.

In summary, there have been a large number of observations from both in vivo and in vitro preparations indicating that changes in LC activity accompany, and participate in producing, changes in behavioral state. There is substantial evidence that this occurs in terms of the sleep–wake cycle, where elevated LC discharge activity precedes spontaneous waking and its associated EEG alerting. Moreover, activating LC with local drug infusions induces neocortical and hippocampal EEG activation, and NE produces discharge patterns characteristic of activated EEG states in individual thalamic and neocortical neurons.

There is also evidence from LC recordings obtained from waking rats, cats, and monkeys that LC activity is modulated within the waking state to produce a more fine-grained control of vigilance. In a parallel set of observations, the application of exogenous NE or LC activation has been found to enhance the robustness and precision of neocortical neuronal responses to defined sensory input while reducing or not altering "background" or "spontaneous" activity. Additionally, NE and/or LC activation enhance several indices of hippocampal synaptic responsiveness, both at the level of individual neurons and at the level of cell populations. Recent evidence indicates that moderate LC activation accompanies optimal information processing, whereas high discharge rates accompany, and perhaps produce, a hyperarousal that may lead to poor performance in circumstances requiring focused, sustained attention.

These latter results suggest that focused attentiveness varies with tonic LC discharge in an inverted U relationship (Aston-Jones, Rajkowski, and Kubiak, in preparation). This relationship between LC activity and performance resembles the Yerkes–Dodson law, suggesting that LC activity changes may in part underlie this classical relationship between "arousal" and performance. It is worthwhile to compare these results for LC activity with those for lesions of the LC system. Lesions of the ascending projections from the LC have led investigators to posit that no or low LC activity may promote attention to contextual, distant cues in the environment, whereas high LC activity (as presumably occurs during stress) may facilitate more focused attention, centered on conditioned or proximal cues (reviewed in Central Norepinephrine Neurons and Behavior). While these lesion results lead to a similar functional dimension for the LC as the cellular recordings (indicating a role in attentional focussing), the specific prediction is the opposite: Recordings predict less focused attention with high LC activity, whereas lesion studies predict more focused attention with high LC activity. These different results may reflect the different species or behavioral paradigms employed. Alternatively, this difference may reveal limitations of the lesion techniques commonly employed in behavioral studies, where weeks are allowed for recovery from surgery before testing and where substantial functional recovery may occur in the lesioned or other systems. More readily interpretable results may be found using acute reversible inactivations of the LC system, where possible recovery of function could not occur (discussed in Central Norepinephrine Neurons and Behavior). Future studies are necessary to test this and other possibilities for the intriguing differences predicted for LC functions from cell recordings as compared to lesion studies.

Despite the continuing progress that characterizes studies of the LC–NE system, some major limitations of the currently available database are readily evident. Four areas ripe for exploitation are the following:

More precise determination of the behavioral correlates of LC discharge. While progress has been made in determining how LC discharge activity is correlated with certain aspects of behavior in well-defined paradigms, much remains to be done. The variety of paradigms in which LC activity has been assessed needs to be expanded, and the number of behavioral, autonomic, and electrographic correlates that are simultaneously indexed and correlated with measures of LC activity must be enlarged.

Moving toward unanesthetized, in vivo preparations in experiments manipulating LC activity. The goal of manipulating LC activity in some specific way and then determining the behavioral consequences of such treatment has remained elusive throughout the history of this field. The experiments described above involving local drug infusions and simultaneous electrophysiological monitoring offer promise in this regard, but they are currently limited to anesthetized preparations. Convincing demonstrations of LC function will depend both on detailed correlative experiments as indicated in the preceding paragraph and on corresponding manipulative experiments that can demonstrate causal relationships between changes in LC discharge activity and indices of behavioral or state variables.

Specifying relationships between LC–NE and other alerting systems. It is almost certain that the LC–NE system performs its functions in such a way that its actions are coordinated with those of other ascending modulatory systems. Systematic study of these interactions will be imperative for understanding the control of forebrain levels of alertness and/or vigilance.

Determining the afferent inputs and efferent targets of LC neurons. The current status of knowledge on this topic is reviewed elsewhere in this volume (see Physiological and Anatomical Determinants of Locus Coeruleus Discharge: Behavior and Clinical Implications). Understanding the functional status of the LC in brain and behavioral activities requires a thorough comprehension of its input–output relationships, as has proved to be so critical in understanding the functions of other brain areas (e.g., sensory, motor). This analysis will reveal the functional correlates of the LC system in brain circuitry, providing important knowledge about the function of the LC system itself.

published 2000